# disable scientific notation
options(scipen = 999)
# dependencies
library(tidyverse)
library(effsize)
library(foreign)
library(timesavers)
library(BayesFactor)
library(weights)
library(sjPlot)
library(nnet)
library(metafor)
library(car)
library(lsr)
library(knitr)
library(kableExtra)data_exp_1 <- read_csv("../experiment 1/data/processed/data_for_analysis.csv")
data_exp_2 <- read_csv("../experiment 2/data/processed/data_for_analysis.csv")
data_exp_3 <- read_csv("../experiment 3/data/processed/data_for_analysis.csv")
data_exp_4 <- read_csv("../experiment 4/data/processed/data_for_analysis.csv")
data_exp_5 <- read_csv("../experiment 5/data/processed/data_for_analysis.csv")
data_exp_6 <- read_csv("../experiment 6/data/processed/data_for_analysis.csv")
data_exp_7 <- read_csv("../experiment 7/data/processed/data_for_analysis.csv")
data_exp_8 <- read_csv("../experiment 8/data/processed/data_for_analysis.csv")
data_exp_9 <- read_csv("../experiment 9/data/processed/data_for_analysis.csv")
# subsets of conditions
data_exp_8nri <- data_exp_8 %>%
filter(Bag_Matching == "Shared_Feature_Bag_Location_Condition_1")
data_exp_8ri <- data_exp_8 %>%
filter(Bag_Matching == "Shared_Feature_Bag_Random_Condition_3")# analysis workflow. is applied witin all studies ---------------
# between groups comparisons analysis workflow ------
t_test_analyses <- function(data) {
# summary stats
summary_stats <- data %>%
group_by(IV) %>%
dplyr::summarize(mean = mean(DV),
sd = sd(DV)) %>%
round_df(2)
# t test
t_test <- t.test(formula = DV ~ IV,
alternative = "two.sided",
paired = FALSE,
data = data)
# effect size
# formula based cohen's d seems to be invariant to ordering of the levels, can't control whether the direction
# cohens_d <- effsize::cohen.d(formula = DV ~ IV,
# paired = FALSE,
# data = data)
cohens_d <- effsize::cohen.d(data$DV[data$IV == "CS1_Positive_CS2_Negative"],
data$DV[data$IV == "CS2_Positive_CS1_Negative"],
paired = FALSE)
# printing
t_test_p <- t_test$p.value[[1]]
t_test_p_apa <- ifelse(t_test_p < 0.001, "< .001",
ifelse(t_test_p < 0.01,
paste("= ", rd(t_test_p, 3), sep = ""), # rd() rounds, converts to string, and removes the leading 0.
paste("= ", rd(t_test_p, 2), sep = "")))
t_test_output <- paste("t(", format(round(t_test$parameter[[1]], 2), nsmall = 2),
") = ", format(round(t_test$statistic[[1]], 2), nsmall = 2),
", p ", t_test_p_apa,
", d = ", format(round(cohens_d$estimate[[1]], 2), nsmall = 2),
", 95% CI = [", format(round(cohens_d$conf.int[["lower"]], 2), nsmall = 2), ", ",
format(round(cohens_d$conf.int[["upper"]], 2), nsmall = 2), "]",
sep = "")
# bayes factor t test
t_test_bf <- ttestBF(formula = DV ~ IV,
data = data)
bf10 <- round(exp(t_test_bf@bayesFactor$bf), 2)
return(list(summary_stats = summary_stats,
t_test = t_test,
cohens_d = cohens_d,
t_test_output = t_test_output,
t_test_p = t_test_p,
t_test_bf = t_test_bf,
bf10 = bf10))
}
# multinominal analysis workflow ------
multinominal_analysis <- function(data) {
# fit model
fit <- multinom(DV ~ IV,
data = data)
# calculate p values
z <- summary(fit)$coefficients/summary(fit)$standard.errors
p <- (1 - pnorm(abs(z), 0, 1)) * 2
pvalues <- p %>%
as.data.frame() %>%
rownames_to_column(var = "var") %>%
rename(p = "IVCS2_Positive_CS1_Negative") %>%
mutate(p = apa_p_value(p)) %>%
select(p)
# calc odds ratios and CIs
coefficients <- as.data.frame(summary(fit)$coefficients) %>%
rownames_to_column(var = "var") %>%
rename(logOR = "IVCS2_Positive_CS1_Negative") %>%
select(var, logOR)
se <- as.data.frame(summary(fit)$standard.errors) %>%
rownames_to_column(var = "var") %>%
rename(logOR_se = "IVCS2_Positive_CS1_Negative") %>%
select(logOR_se)
# combine
results <- cbind(coefficients, se) %>%
mutate(OR = round(exp(logOR), 2),
OR_ci_lwr = round(exp(logOR - logOR_se*1.96), 2),
OR_ci_upr = round(exp(logOR + logOR_se*1.96), 2)) %>%
cbind(pvalues)
return(results)
}
# meta analysis workflow ------
add_heterogeneity_metrics_to_forest <- function(fit) {
bquote(paste("RE Model (",
italic('I')^"2", " = ", .(formatC(format(round(fit$I2, 1), nsmall = 1))),
"%, ", italic('H')^"2", " = ", .(formatC(format(round(fit$H2, 1), nsmall = 1))), ")"))
}
meta_analysis_workflow <- function(data,
effect_size_label,
reference_line,
exp_estimates = FALSE,
plot = TRUE) {
require(timesavers)
require(tidyverse)
require(metafor)
# fit random effects model
fitted_model <-
rma(yi = yi,
sei = sei,
data = data,
slab = Experiment)
p_value <- apa_p_value(fitted_model$pval)
# model predictions
meta_analysis_results <-
predict(fitted_model, digits = 5) %>%
as.data.frame() %>%
gather() %>%
round_df(2) %>%
dplyr::rename(metric = key,
estimate = value) %>%
mutate(metric = dplyr::recode(metric,
"pred" = paste0("Meta analysed ", effect_size_label),
"ci.lb" = "95% CI lower",
"ci.ub" = "95% CI upper",
"cr.lb" = "95% CR lower",
"cr.ub" = "95% CR upper"))
# source_valence_conditionally exponentiate estimates if converting log odds to odds ratios
if(exp_estimates == TRUE) {
meta_analysis_results <- meta_analysis_results %>%
mutate(estimate = round(exp(estimate), 2))
}
meta_analysis_results <- rbind(meta_analysis_results,
data.frame(metric = "p",
estimate = p_value))
# summarize results
meta_analysis_results_text <-
paste0("k = ", fitted_model$k, ", ",
effect_size_label, " = ", meta_analysis_results$estimate[1],
# dynamic indexing of some values as different models return variables in different locations, but relative location is reliable
", 95% CI = [", meta_analysis_results$estimate[length(meta_analysis_results$estimate)-4],
", ", meta_analysis_results$estimate[length(meta_analysis_results$estimate)-3],
"], 95% CR = [", meta_analysis_results$estimate[length(meta_analysis_results$estimate)-2],
", ", meta_analysis_results$estimate[length(meta_analysis_results$estimate)-1],
"], p ", meta_analysis_results$estimate[length(meta_analysis_results$estimate)])
heterogeneity_test_results_text <-
paste0("Q(df = ", fitted_model$k - 1, ") = ", round(fitted_model$QE, 2),
", p ", ifelse(fitted_model$pval < 0.001, "< .001", as.character(paste("=", round(fitted_model$pval, 3)))),
", tau^2 = ", round(fitted_model$tau2, 2),
", I^2 = ", round(fitted_model$I2, 2), "%",
", H^2 = ", round(fitted_model$H2, 2))
# forest plot
if (exp_estimates == TRUE) {
transformation <- exp
} else {
transformation <- NULL
}
if (plot == TRUE) {
forest_plot <- forest(fitted_model,
xlab = effect_size_label,
addcred = TRUE,
atransf = transformation,
refline = reference_line,
mlab = add_heterogeneity_metrics_to_forest(fitted_model))
} else {
forest_plot <- NULL
}
return(list(fitted_model = fitted_model,
meta_analysis_results = meta_analysis_results,
meta_analysis_results_text = meta_analysis_results_text,
heterogeneity_test_results_text = heterogeneity_test_results_text,
forest_plot = forest_plot))
}
# robustness between original test and subset -----------
robust_test <- function(subset_test, original_test) {
subset_p <- subset_test$estimate[subset_test$metric == "p"] %>%
str_remove(., "=") %>%
str_remove(., "<") %>%
str_remove(., " ") %>%
as.numeric()
original_p <- original_test$estimate[original_test$metric == "p"] %>%
str_remove(., "=") %>%
str_remove(., "<") %>%
str_remove(., " ") %>%
as.numeric()
return((subset_p >= 0.05 & original_p >= 0.05) | (subset_p < 0.05 & original_p < 0.05))
}
# moderator meta analysis workflow ------
multilevel_moderator_meta_analysis_workflow <- function(data,
effect_size_label = "Cohen's d") {
require(timesavers)
require(tidyverse)
require(metafor)
# fit random effects model
fitted_model <-
rma.mv(yi = yi,
V = sei^2,
mods = ~ exclude,
random = ~as.factor(exclude) | Experiment,
data = data,
slab = paste(Experiment, ifelse(exclude == TRUE, "excluded", "retained"), sep = " - "))
# summarize moderator results
mod_meta_analysis_results_text <-
paste0("QM(df = 1) = ", round(fitted_model$QM, 2),
", p ", ifelse(fitted_model$QMp < 0.001, "< .001", as.character(paste("=", round(fitted_model$QMp, 3)))),
", difference in excluded subset ", effect_size_label, " = ", round(fitted_model$b[2], 2),
", 95% CI [", round(fitted_model$ci.lb[2], 2), ", ", round(fitted_model$ci.ub[2], 2), "]")
return(list(fitted_model = fitted_model,
mod_meta_analysis_results_text = mod_meta_analysis_results_text))
}data_exp_1 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 103 | 33.06 | 8.43 |
data_exp_1 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| Female | 57 |
| Male | 46 |
results_exp_1_h1 <- data_exp_1 %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.37, SD = 0.46), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -0.23, SD = 0.45), t(98.12) = 6.63, p < .001, d = 1.31, 95% CI = [0.88, 1.74], BF10 = 5714391.64.
results_exp_1_h2 <- data_exp_1 %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 3.33, SD = 4.60), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -4.15, SD = 4.48), t(98.32) = 8.33, p < .001, d = 1.65, 95% CI = [1.20, 2.10], BF10 = 15170767375.77.
# set reference level to "Morag"
data_exp_1$intention <- relevel(as.factor(data_exp_1$intention), ref = "Morag")
results_exp_1_h3 <- data_exp_1 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, congruent with training, OR = 13.66, 95% CI = [3.56, 52.44], p = .0001.
data_exp_2 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 103 | 32.5 | 8.87 |
data_exp_2 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| Female | 58 |
| Male | 45 |
results_exp_2_h1 <- data_exp_2 %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS2 eventually shared a color with positive words and CS1 shared a color with negative words, participants showed an automatic relative preference for CS2 over CS1 on the IAT (M = -0.26, SD = 0.54). whereas when CS2 eventually shared a color with negative words and CS1 shared a color with positive words, they demonstrated a relative preference for CS1 over CS2 (M = -0.12, SD = 0.60), t(100.85) = -1.18, p = .24, d = -0.23, 95% CI = [-0.62, 0.16], BF10 = 0.38.
Note that this result is in the opposite direction to all other experiments.
results_exp_2_h2 <- data_exp_2 %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)Regardless of whether CS1 eventually shared a color with positive or negative words, participants showed a relative preference on the self-report ratings for CS2 over CS1 (CS1-positive condition: M = -2.80, SD = 5.33; CS1-negative condition: M = -1.58, SD = 6.03), t(100.98) = -1.09, p = .28, d = -0.21, 95% CI = [-0.61, 0.18], BF10 = 0.35.
Note that this null result is different to all other experiments.
# set reference level to "Morag"
data_exp_2$intention <- relevel(as.factor(data_exp_2$intention), ref = "Morag")
results_exp_2_h3 <- data_exp_2 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, but in the opposite direction to that predicted, OR = 0.22, 95% CI = [0.07, 0.66], p = .0073.
Note that this result is in the opposite direction to all other experiments.
Experiment 3 was informed by the unexpected results of experiment 2, which prompted us to examine the distribution of scores on the IAT and self reports which suggested bi- and tri-modality. This is depicted below.
p1 <-
data_exp_2 %>%
select(Participant, source_valence_condition, IAT_Score, EC_Effect_CS1_CS2_Difference_Score) %>%
mutate(source_valence_condition = dplyr::recode(source_valence_condition,
CS1_Positive_CS2_Negative = "Morag positive & Struan negative",
CS2_Positive_CS1_Negative = "Morag negative & Struan positive")) %>%
rename(`IAT D score` = IAT_Score,
`Ratings difference score` = EC_Effect_CS1_CS2_Difference_Score) %>%
gather(dv_modality, dv_score, c(`IAT D score`, `Ratings difference score`)) %>%
ggplot(aes(dv_score)) +
geom_density(adjust = 0.5) +
facet_wrap(~dv_modality+source_valence_condition, scales = "free") +
theme_classic() +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
strip.background = element_blank(),
panel.border = element_blank()) +
xlab("") +
ylab("Probability density")
p1data_exp_3 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 97 | 28.67 | 6.07 |
data_exp_3 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| Female | 58 |
| Male | 39 |
results_exp_3_h1 <- data_exp_3 %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.21, SD = 0.46), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -0.15, SD = 0.59), t(93.42) = 3.29, p = .001, d = 0.66, 95% CI = [0.25, 1.08], BF10 = 20.10.
results_exp_3_h2 <- data_exp_3 %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 2.92, SD = 5.25), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -2.85, SD = 5.02), t(92.94) = 5.52, p < .001, d = 1.13, 95% CI = [0.69, 1.56], BF10 = 45262.88.
# set reference level to "Morag"
data_exp_3$intention <- relevel(as.factor(data_exp_3$intention), ref = "Morag")
results_exp_3_h3 <- data_exp_3 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, congruent with training, OR = 6.94, 95% CI = [2.03, 23.77], p = .002.
data_exp_4 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 187 | 30.3 | 6.2 |
data_exp_4 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| Female | 94 |
| Male | 93 |
results_exp_4_h1 <- data_exp_4 %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.16, SD = 0.48), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -0.18, SD = 0.46), t(184.00) = 5.02, p < .001, d = 0.73, 95% CI = [0.44, 1.03], BF10 = 11902.22.
results_exp_4_h2 <- data_exp_4 %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 3.57, SD = 4.99), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -2.18, SD = 4.21), t(179.27) = 8.51, p < .001, d = 1.25, 95% CI = [0.93, 1.56], BF10 = 922634733581.12.
# set reference level to "Morag"
data_exp_4$intention <- relevel(as.factor(data_exp_4$intention), ref = "Morag")
results_exp_4_h3 <- data_exp_4 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, congruent with training, OR = 7.63, 95% CI = [3.11, 18.76], p < .0001.
data_exp_5 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 179 | 30.92 | 5.85 |
data_exp_5 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| Female | 95 |
| Male | 84 |
results_exp_5_h1 <- data_exp_5 %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.14, SD = 0.46), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -0.12, SD = 0.46), t(168.75) = 3.79, p < .001, d = 0.57, 95% CI = [0.27, 0.87], BF10 = 109.05.
results_exp_5_h2 <- data_exp_5 %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 2.34, SD = 4.12), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -2.38, SD = 4.09), t(169.77) = 7.66, p < .001, d = 1.15, 95% CI = [0.83, 1.47], BF10 = 5415593221.07.
# set reference level to "Morag"
data_exp_5$intention <- relevel(as.factor(data_exp_5$intention), ref = "Morag")
results_exp_5_h3 <- data_exp_5 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, congruent with training, OR = 5.00, 95% CI = [1.91, 13.06], p = .001.
data_exp_6 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 231 | 27.34 | 7.33 |
data_exp_6 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| female | 103 |
| male | 125 |
| other | 3 |
results_exp_6_h1 <- data_exp_6 %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.31, SD = 0.51), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -0.18, SD = 0.52), t(227.66) = 7.25, p < .001, d = 0.96, 95% CI = [0.68, 1.23], BF10 = 1151492207.55.
results_exp_6_h2 <- data_exp_6 %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 3.61, SD = 4.72), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -3.40, SD = 4.09), t(227.44) = 12.08, p < .001, d = 1.58, 95% CI = [1.29, 1.88], BF10 = 72613291696428081479680.00.
# set reference level to "Morag"
data_exp_6$intention <- relevel(as.factor(data_exp_6$intention), ref = "Morag")
results_exp_6_h3 <- data_exp_6 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, congruent with training, OR = 14.85, 95% CI = [6.98, 31.63], p < .0001.
data_exp_7 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age, na.rm = TRUE),
age_sd = sd(Age, na.rm = TRUE)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 493 | 28.28 | 7.43 |
data_exp_7 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| female | 242 |
| male | 246 |
| NA | 5 |
These analyses code each trial as being congruent or incongruent based on the valence of the prime and the valence of the source stimulus valence. Mean reaction time was calculated for each participant’s congruent and incongruent trials. These means for each trial type are then compared between participants. This represents the classical test of the EP effect.
data_300 <- data_exp_7 %>%
slice(1:300) %>%
select(ep_mean_rt_consistent, ep_mean_rt_inconsistent) %>%
gather()
ttestBF(x = data_300$value[data_300$key == "ep_mean_rt_inconsistent"],
y = data_300$value[data_300$key == "ep_mean_rt_consistent"],
paired = TRUE)## Bayes factor analysis
## --------------
## [1] Alt., r=0.707 : 1.41985 ±0%
##
## Against denominator:
## Null, mu = 0
## ---
## Bayes factor type: BFoneSample, JZS
sample_300 <-
ttestBF(x = data_300$value[data_300$key == "ep_mean_rt_inconsistent"],
y = data_300$value[data_300$key == "ep_mean_rt_consistent"],
paired = TRUE,
posterior = TRUE,
iterations = 5000)
data.frame(sample_300) %>%
dplyr::summarize(mu_median = quantile(mu, prob = .500),
mu_lower = quantile(mu, prob = .025),
mu_upper = quantile(mu, prob = .975),
delta_median = quantile(delta, prob = .500),
delta_lower = quantile(delta, prob = .025),
delta_upper = quantile(delta, prob = .975)) %>%
round(2) %>%
gather(parameter, estimate)## parameter estimate
## 1 mu_median 4.20
## 2 mu_lower 0.95
## 3 mu_upper 7.50
## 4 delta_median 0.14
## 5 delta_lower 0.03
## 6 delta_upper 0.25
data_400 <- data_exp_7 %>%
slice(1:400) %>%
select(ep_mean_rt_consistent, ep_mean_rt_inconsistent) %>%
gather()
ttestBF(x = data_400$value[data_400$key == "ep_mean_rt_inconsistent"],
y = data_400$value[data_400$key == "ep_mean_rt_consistent"],
paired = TRUE)## Bayes factor analysis
## --------------
## [1] Alt., r=0.707 : 0.7645194 ±0.01%
##
## Against denominator:
## Null, mu = 0
## ---
## Bayes factor type: BFoneSample, JZS
sample_400 <-
ttestBF(x = data_400$value[data_400$key == "ep_mean_rt_inconsistent"],
y = data_400$value[data_400$key == "ep_mean_rt_consistent"],
paired = TRUE,
posterior = TRUE,
iterations = 5000)
data.frame(sample_400) %>%
dplyr::summarize(mu_median = quantile(mu, prob = .500),
mu_lower = quantile(mu, prob = .025),
mu_upper = quantile(mu, prob = .975),
delta_median = quantile(delta, prob = .500),
delta_lower = quantile(delta, prob = .025),
delta_upper = quantile(delta, prob = .975)) %>%
round(2) %>%
gather(parameter, estimate)## parameter estimate
## 1 mu_median 3.45
## 2 mu_lower 0.48
## 3 mu_upper 6.30
## 4 delta_median 0.12
## 5 delta_lower 0.02
## 6 delta_upper 0.21
data_500 <- data_exp_7 %>%
select(ep_mean_rt_consistent, ep_mean_rt_inconsistent) %>%
gather()
ttestBF(x = data_500$value[data_500$key == "ep_mean_rt_inconsistent"],
y = data_500$value[data_500$key == "ep_mean_rt_consistent"],
paired = TRUE)## Bayes factor analysis
## --------------
## [1] Alt., r=0.707 : 3.113847 ±0%
##
## Against denominator:
## Null, mu = 0
## ---
## Bayes factor type: BFoneSample, JZS
sample_500 <-
ttestBF(x = data_500$value[data_500$key == "ep_mean_rt_inconsistent"],
y = data_500$value[data_500$key == "ep_mean_rt_consistent"],
paired = TRUE,
posterior = TRUE,
iterations = 5000)
data.frame(sample_500) %>%
dplyr::summarize(mu_median = quantile(mu, prob = .500),
mu_lower = quantile(mu, prob = .025),
mu_upper = quantile(mu, prob = .975),
delta_median = quantile(delta, prob = .500),
delta_lower = quantile(delta, prob = .025),
delta_upper = quantile(delta, prob = .975)) %>%
round(2) %>%
gather(parameter, estimate)## parameter estimate
## 1 mu_median 3.88
## 2 mu_lower 1.22
## 3 mu_upper 6.40
## 4 delta_median 0.13
## 5 delta_lower 0.04
## 6 delta_upper 0.21
Frequentist analysis of the final sample for reader familiarity
t.test(x = data_500$value[data_500$key == "ep_mean_rt_inconsistent"],
y = data_500$value[data_500$key == "ep_mean_rt_consistent"],
paired = TRUE)##
## Paired t-test
##
## data: data_500$value[data_500$key == "ep_mean_rt_inconsistent"] and data_500$value[data_500$key == "ep_mean_rt_consistent"]
## t = 2.8909, df = 492, p-value = 0.004011
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 1.245710 6.531166
## sample estimates:
## mean of the differences
## 3.888438
##
## Cohen's d
##
## d estimate: 0.04949019 (negligible)
## 95 percent confidence interval:
## lower upper
## 0.01587567 0.08310471
Using the final analytic sample.
results_exp_7_h2 <- data_exp_7 %>%
select(EC_Effect_CS1_CS2_Difference_Score, source_valence_condition) %>%
na.omit() %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 2.17, SD = 4.35), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = -2.66, SD = 4.02), t(479.98) = 12.72, p < .001, d = 1.15, 95% CI = [0.96, 1.35], BF10 = 80819093185460097270287433728.00.
Using the final analytic sample.
# set reference level to "Morag"
data_exp_7$intention <- relevel(as.factor(data_exp_7$intention), ref = "Morag")
results_exp_7_h3 <- data_exp_7 %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()Data from the behavioural intentions question were entered into a multinomial logistic regression with CSI as the reference category. Only results from the CS1-CS2 comparison are relevant to the hypothesis and will be reported here. Results demonstrated that participants responses of CS1 relative to CS2 differed between the two conditions, congruent with training, OR = 4.33, 95% CI = [2.77, 6.78], p < .0001.
data_exp_8 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 222 | 28.84 | 7.54 |
data_exp_8 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| female | 127 |
| male | 92 |
| other | 3 |
Pairwise follow up t-tests using Bonferroni-Holm corrections for multiple testing.
Results are not parsed into text automatically given the analysis workflow differs from other experiments, and this would take more coding effort than its worth.
# anova
fit_exp_8_h1 <- aov(IAT_Score ~ Bag_Matching,
contrasts = "contr.sum",
data = data_exp_8)
Anova(fit_exp_8_h1, type = "III")## Anova Table (Type III tests)
##
## Response: IAT_Score
## Sum Sq Df F value Pr(>F)
## (Intercept) 3.432 1 19.854 0.00001333305091 ***
## Bag_Matching 9.770 2 28.261 0.00000000001206 ***
## Residuals 37.854 219
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## eta.sq eta.sq.part
## Bag_Matching 0.2051465 0.2051465
##
## Pairwise comparisons using t tests with pooled SD
##
## data: data_exp_8$IAT_Score and data_exp_8$Bag_Matching
##
## No_Shared_Feature_Condition_2
## Shared_Feature_Bag_Location_Condition_1 0.000000000041
## Shared_Feature_Bag_Random_Condition_3 0.000000515586
## Shared_Feature_Bag_Location_Condition_1
## Shared_Feature_Bag_Location_Condition_1 -
## Shared_Feature_Bag_Random_Condition_3 0.1
##
## P value adjustment method: holm
# means by condition
data_exp_8 %>%
group_by(Bag_Matching) %>%
dplyr::summarize(mean = mean(IAT_Score),
ci_lwr = mean - plotrix::std.error(IAT_Score)*1.96,
ci_upr = mean + plotrix::std.error(IAT_Score)*1.96) %>%
round_df(2)## # A tibble: 3 x 4
## Bag_Matching mean ci_lwr ci_upr
## <chr> <dbl> <dbl> <dbl>
## 1 No_Shared_Feature_Condition_2 -0.2 -0.290 -0.12
## 2 Shared_Feature_Bag_Location_Condition_1 0.28 0.18 0.38
## 3 Shared_Feature_Bag_Random_Condition_3 0.16 0.06 0.26
# anova
fit_exp_8_h2 <- aov(EC_Effect_CS1_CS2_Difference_Score ~ Bag_Matching,
contrasts = "contr.sum",
data = data_exp_8)
Anova(fit_exp_8_h2, type = "III")## Anova Table (Type III tests)
##
## Response: EC_Effect_CS1_CS2_Difference_Score
## Sum Sq Df F value Pr(>F)
## (Intercept) 109.14 1 8.9765 0.00305 **
## Bag_Matching 649.36 2 26.7035 0.0000000000419 ***
## Residuals 2662.77 219
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## eta.sq eta.sq.part
## Bag_Matching 0.196056 0.196056
# follow up t tests
pairwise.t.test(data_exp_8$EC_Effect_CS1_CS2_Difference_Score,
data_exp_8$Bag_Matching,
p.adj = "holm")##
## Pairwise comparisons using t tests with pooled SD
##
## data: data_exp_8$EC_Effect_CS1_CS2_Difference_Score and data_exp_8$Bag_Matching
##
## No_Shared_Feature_Condition_2
## Shared_Feature_Bag_Location_Condition_1 0.00000000062
## Shared_Feature_Bag_Random_Condition_3 0.00000008789
## Shared_Feature_Bag_Location_Condition_1
## Shared_Feature_Bag_Location_Condition_1 -
## Shared_Feature_Bag_Random_Condition_3 0.39
##
## P value adjustment method: holm
# means by condition
data_exp_8 %>%
group_by(Bag_Matching) %>%
dplyr::summarize(mean = mean(EC_Effect_CS1_CS2_Difference_Score),
ci_lwr = mean - plotrix::std.error(EC_Effect_CS1_CS2_Difference_Score)*1.96,
ci_upr = mean + plotrix::std.error(EC_Effect_CS1_CS2_Difference_Score)*1.96) %>%
round_df(2)## # A tibble: 3 x 4
## Bag_Matching mean ci_lwr ci_upr
## <chr> <dbl> <dbl> <dbl>
## 1 No_Shared_Feature_Condition_2 -1.14 -1.78 -0.5
## 2 Shared_Feature_Bag_Location_Condition_1 2.61 1.73 3.49
## 3 Shared_Feature_Bag_Random_Condition_3 2.1 1.21 3
results_exp_8nri_h1 <- data_exp_8nri %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.44, SD = 0.32), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = 0.10, SD = 0.46), t(58.42) = 3.50, p < .001, d = 0.84, 95% CI = [0.35, 1.34], BF10 = 42.20.
results_exp_8nri_h2 <- data_exp_8nri %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 3.57, SD = 2.82), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = 1.57, SD = 4.41), t(55.35) = 2.26, p = .03, d = 0.55, 95% CI = [0.06, 1.03], BF10 = 2.25.
results_exp_8ri_h1 <- data_exp_8ri %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed an automatic relative preference for CS1 over CS2 on the IAT (M = 0.45, SD = 0.40), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = 0.01, SD = 0.35), t(40.42) = 4.53, p < .001, d = 1.21, 95% CI = [0.66, 1.77], BF10 = 1255.98.
results_exp_8ri_h2 <- data_exp_8ri %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)When CS1 eventually shared a color with positive words and CS2 shared a color with negative words, participants showed a relative preference on the self-report ratings for CS1 over CS2 (M = 3.53, SD = 3.90), whereas when CS1 eventually shared a color with negative words and CS2 shared a color with positive words, they demonstrated a relative preference for CS2 over CS1 (M = 1.35, SD = 3.46), t(40.33) = 2.26, p = .03, d = 0.60, 95% CI = [0.08, 1.13], BF10 = 2.52.
data_exp_9 %>%
dplyr::summarize(n = n(),
age_mean = mean(Age),
age_sd = sd(Age)) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| n | age_mean | age_sd |
|---|---|---|
| 89 | 31.53 | 7.35 |
data_exp_9 %>%
dplyr::count(Gender) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| Gender | n |
|---|---|
| female | 47 |
| male | 42 |
##
## One Sample t-test
##
## data: data_exp_9$IAT_Score
## t = -3.8952, df = 88, p-value = 0.0001908
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## -0.24189282 -0.07845558
## sample estimates:
## mean of x
## -0.1601742
## # A tibble: 1 x 1
## d
## <dbl>
## 1 -0.41
##
## One Sample t-test
##
## data: data_exp_9$EC_Effect_CS1_CS2_Difference_Score
## t = -5.7116, df = 88, p-value = 0.0000001495
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## -2.101422 -1.016555
## sample estimates:
## mean of x
## -1.558989
data_exp_9 %>%
dplyr::summarize(d = round(mean(EC_Effect_CS1_CS2_Difference_Score)/sd(EC_Effect_CS1_CS2_Difference_Score), 2))## # A tibble: 1 x 1
## d
## <dbl>
## 1 -0.61
Meta analyses of experiments 1-8 (nb not 9, which contained no manipulations but acted as an extension of experiment 8).
Random effects meta analyses were fitted using the metafor R package (Viechtbauer, 2010). The maximum liklihood estimator was used.
A separate meta analysis was fitted for each outcome variable (IAT, self reported ratings, behavioural intentions). Although multivariate meta analysis was a possible alternative (and is conducted in a separate markdown script), previous work has traditionally assessed these dependant variables separately, so we will too.
Note that not all variables were collected in all subsets, so sample sizes for subsets do not equal the total sample size * the proportion meeting a given criteria.
total_n_across_studies <-
nrow(data_exp_1) +
nrow(data_exp_2) +
nrow(data_exp_3) +
nrow(data_exp_4) +
nrow(data_exp_5) +
nrow(data_exp_6) +
nrow(data_exp_7) +
nrow(data_exp_8)
data_combined <-
bind_rows(mutate(data_exp_1, experiment = 1),
mutate(data_exp_2, experiment = 2),
mutate(data_exp_3, experiment = 3),
mutate(data_exp_4, experiment = 4),
mutate(data_exp_5, experiment = 5),
mutate(data_exp_6, experiment = 6),
mutate(data_exp_7, experiment = 7),
mutate(data_exp_8nri, experiment = 8),
mutate(data_exp_8ri, experiment = 8)) %>%
select(experiment, US_contingency_memory, color_valence_contingency_memory,
demand_IAT, demand_self_reports, hypothesis_awareness, reactance_IAT, reactance_self_reports) %>%
# fudge to stop "unknown variable" warning from appearing
mutate(DV = NA)
data_combined %>%
dplyr::summarize(full_sample_n = n()) %>%
round_df(2) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| full_sample_n |
|---|
| 1531 |
# percents by subset
temp_1 <- data_combined %>%
count(US_contingency_memory) %>%
spread(US_contingency_memory, n) %>%
mutate(US_contingency_aware = `TRUE` / (`TRUE` + `FALSE`)) %>%
select(US_contingency_aware)
temp_2 <- data_combined %>%
count(color_valence_contingency_memory) %>%
spread(color_valence_contingency_memory, n) %>%
mutate(color_valence_contingency_aware = `TRUE` / (`TRUE` + `FALSE`)) %>%
select(color_valence_contingency_aware)
temp_3 <- data_combined %>%
rowwise() %>%
mutate(demand = as.logical(max(c(demand_IAT, demand_self_reports)))) %>%
ungroup() %>%
count(demand) %>%
spread(demand, n) %>%
mutate(NOT_demand_compliant = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(NOT_demand_compliant)
temp_4a <- data_combined %>%
count(reactance_IAT) %>%
spread(reactance_IAT, n) %>%
mutate(NOT_reactant_IAT = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(NOT_reactant_IAT)
temp_4b <- data_combined %>%
count(reactance_self_reports) %>%
spread(reactance_self_reports, n) %>%
mutate(NOT_reactant_self_reports = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(NOT_reactant_self_reports)
temp_5 <- data_combined %>%
count(hypothesis_awareness) %>%
spread(hypothesis_awareness, n) %>%
mutate(NOT_hypothesis_awareness = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(NOT_hypothesis_awareness)
bind_cols(temp_1,
temp_2,
temp_3,
temp_4a,
temp_4b,
temp_5) %>%
gather(subset, proportion) %>%
round_df(3) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| subset | proportion |
|---|---|
| US_contingency_aware | 0.775 |
| color_valence_contingency_aware | 0.596 |
| NOT_demand_compliant | 0.809 |
| NOT_reactant_IAT | 0.897 |
| NOT_reactant_self_reports | 0.858 |
| NOT_hypothesis_awareness | 0.267 |
# percents by subset
temp_1_by_exp <- data_combined %>%
count(experiment, US_contingency_memory) %>%
spread(US_contingency_memory, n) %>%
mutate(US_contingency_aware = `TRUE` / (`TRUE` + `FALSE`)) %>%
select(experiment, US_contingency_aware)
temp_2_by_exp <- data_combined %>%
count(experiment, color_valence_contingency_memory) %>%
spread(color_valence_contingency_memory, n) %>%
mutate(color_valence_contingency_aware = `TRUE` / (`TRUE` + `FALSE`)) %>%
select(experiment, color_valence_contingency_aware)
temp_3a_by_exp <- data_combined %>%
count(experiment, demand_IAT) %>%
spread(demand_IAT, n) %>%
mutate(NOT_demand_IAT = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(experiment, NOT_demand_IAT)
temp_3b_by_exp <- data_combined %>%
count(experiment, demand_self_reports) %>%
spread(demand_self_reports, n) %>%
mutate(NOT_demand_self_reports = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(experiment, NOT_demand_self_reports)
temp_4a_by_exp <- data_combined %>%
count(experiment, reactance_IAT) %>%
spread(reactance_IAT, n) %>%
mutate(NOT_reactant_IAT = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(experiment, NOT_reactant_IAT)
temp_4b_by_exp <- data_combined %>%
count(experiment, reactance_self_reports) %>%
spread(reactance_self_reports, n) %>%
mutate(NOT_reactant_self_reports = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(experiment, NOT_reactant_self_reports)
temp_5_by_exp <- data_combined %>%
count(experiment, hypothesis_awareness) %>%
spread(hypothesis_awareness, n) %>%
mutate(NOT_hypothesis_awareness = `FALSE` / (`TRUE` + `FALSE`)) %>%
select(experiment, NOT_hypothesis_awareness)
temp_1_by_exp %>%
left_join(temp_2_by_exp, by = "experiment") %>%
left_join(temp_3a_by_exp, by = "experiment") %>%
left_join(temp_3b_by_exp, by = "experiment") %>%
left_join(temp_4a_by_exp, by = "experiment") %>%
left_join(temp_4b_by_exp, by = "experiment") %>%
left_join(temp_5_by_exp, by = "experiment") %>%
round_df(3) %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| experiment | US_contingency_aware | color_valence_contingency_aware | NOT_demand_IAT | NOT_demand_self_reports | NOT_reactant_IAT | NOT_reactant_self_reports | NOT_hypothesis_awareness |
|---|---|---|---|---|---|---|---|
| 1 | 0.650 | 0.718 | 0.825 | 0.845 | 0.806 | 0.835 | 0.282 |
| 2 | 0.650 | 0.621 | 0.854 | 0.883 | 0.903 | 0.845 | 0.184 |
| 3 | 0.794 | 0.794 | 0.814 | 0.773 | 0.887 | 0.856 | 0.082 |
| 4 | 0.781 | 0.620 | 0.925 | 0.824 | 0.904 | 0.850 | 0.326 |
| 5 | 0.749 | 0.453 | 0.860 | 0.804 | 0.916 | 0.849 | 0.140 |
| 6 | NA | NA | 0.983 | 0.918 | 0.926 | 0.848 | 0.329 |
| 7 | 0.781 | 0.568 | 0.930 | 0.859 | 0.895 | 0.900 | 0.348 |
| 8 | 0.949 | NA | 0.986 | 0.870 | 0.891 | 0.775 | 0.145 |
Total N in all studies (for reporting in abstract) = 1615.
Notes on meta analysed effect size:
Credibility intervals refer to the interval of effect sizes likely to be observed in a future study on the basis of the width of estimation of the true effect size combined with variation in this true effect size (e.g., due to heterogeneity).
Notes on heterogeneity metrics:
results_for_meta_analysis_h1 <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1$cohens_d$estimate[[1]],
results_exp_2_h1$cohens_d$estimate[[1]],
results_exp_3_h1$cohens_d$estimate[[1]],
results_exp_4_h1$cohens_d$estimate[[1]],
results_exp_5_h1$cohens_d$estimate[[1]],
results_exp_6_h1$cohens_d$estimate[[1]],
results_exp_8nri_h1$cohens_d$estimate[[1]],
results_exp_8ri_h1$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1$cohens_d$conf.int[["lower"]],
results_exp_2_h1$cohens_d$conf.int[["lower"]],
results_exp_3_h1$cohens_d$conf.int[["lower"]],
results_exp_4_h1$cohens_d$conf.int[["lower"]],
results_exp_5_h1$cohens_d$conf.int[["lower"]],
results_exp_6_h1$cohens_d$conf.int[["lower"]],
results_exp_8nri_h1$cohens_d$conf.int[["lower"]],
results_exp_8ri_h1$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1$cohens_d$conf.int[["upper"]],
results_exp_2_h1$cohens_d$conf.int[["upper"]],
results_exp_3_h1$cohens_d$conf.int[["upper"]],
results_exp_4_h1$cohens_d$conf.int[["upper"]],
results_exp_5_h1$cohens_d$conf.int[["upper"]],
results_exp_6_h1$cohens_d$conf.int[["upper"]],
results_exp_8nri_h1$cohens_d$conf.int[["upper"]],
results_exp_8ri_h1$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2))
results_for_meta_analysis_h1 %>%
select(Experiment, yi, sei) %>%
write_csv("../meta analysis data/results_for_meta_analysis_iat.csv")
meta_results_h1 <- meta_analysis_workflow(data = results_for_meta_analysis_h1,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0)Undue influence of one or more studies on either the meta analysed effect size or the heterogeneity among studies was then assessed.
Note on influence metrics from the metafor package documentation:
Outliers were assessed on the baiss of undue influence, on the predicted average effect size/coefficients (DFFITS), and the large decrease in heterogeneity on the basis of its removal (tau2.del, QE.del). Note that outliers were also selected based on consistency across all three DVs (see other analyses below).
influence_meta_results_h1 <- influence(meta_results_h1$fitted_model)
print(influence_meta_results_h1, digits = 2) ##
## rstudent dffits cook.d cov.r tau2.del QE.del hat
## Experiment 1 1.35 0.51 0.23 1.01 0.15 29.06 0.12
## Experiment 2 -3.84 -1.25 0.51 0.36 0.03 11.24 0.13
## Experiment 3 -0.18 -0.08 0.01 1.33 0.21 36.70 0.12
## Experiment 4 -0.03 -0.02 0.00 1.37 0.22 36.81 0.14
## Experiment 5 -0.39 -0.16 0.03 1.34 0.21 35.54 0.14
## Experiment 6 0.47 0.18 0.04 1.32 0.21 33.41 0.14
## Experiment 8 (standard) 0.20 0.06 0.00 1.30 0.21 36.58 0.11
## Experiment 8 (random) 0.97 0.33 0.11 1.13 0.18 33.74 0.11
## weight dfbs inf
## Experiment 1 12.10 0.51
## Experiment 2 12.56 -1.27 *
## Experiment 3 12.30 -0.08
## Experiment 4 13.63 -0.02
## Experiment 5 13.58 -0.16
## Experiment 6 13.88 0.18
## Experiment 8 (standard) 11.33 0.06
## Experiment 8 (random) 10.61 0.33
Experiment 2 was assessed as being an outlier.
A sensitivity analysis therefore excluded Experiment 2 and re meta analysed.
results_for_meta_analysis_h1_sensitivity <- results_for_meta_analysis_h1 %>%
filter(Experiment != "Experiment 2")
meta_results_h1_sensitivity <-
meta_analysis_workflow(data = results_for_meta_analysis_h1_sensitivity,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0)results_for_meta_analysis_h2 <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 7",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2$cohens_d$estimate[[1]],
results_exp_2_h2$cohens_d$estimate[[1]],
results_exp_3_h2$cohens_d$estimate[[1]],
results_exp_4_h2$cohens_d$estimate[[1]],
results_exp_5_h2$cohens_d$estimate[[1]],
results_exp_6_h2$cohens_d$estimate[[1]],
results_exp_7_h2$cohens_d$estimate[[1]],
results_exp_8nri_h2$cohens_d$estimate[[1]],
results_exp_8ri_h2$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2$cohens_d$conf.int[["lower"]],
results_exp_2_h2$cohens_d$conf.int[["lower"]],
results_exp_3_h2$cohens_d$conf.int[["lower"]],
results_exp_4_h2$cohens_d$conf.int[["lower"]],
results_exp_5_h2$cohens_d$conf.int[["lower"]],
results_exp_6_h2$cohens_d$conf.int[["lower"]],
results_exp_7_h2$cohens_d$conf.int[["lower"]],
results_exp_8nri_h2$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2$cohens_d$conf.int[["upper"]],
results_exp_2_h2$cohens_d$conf.int[["upper"]],
results_exp_3_h2$cohens_d$conf.int[["upper"]],
results_exp_4_h2$cohens_d$conf.int[["upper"]],
results_exp_5_h2$cohens_d$conf.int[["upper"]],
results_exp_6_h2$cohens_d$conf.int[["upper"]],
results_exp_7_h2$cohens_d$conf.int[["upper"]],
results_exp_8nri_h2$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2))
results_for_meta_analysis_h2 %>%
select(Experiment, yi, sei) %>%
write_csv("../meta analysis data/results_for_meta_analysis_self_report.csv")
meta_results_h2 <- meta_analysis_workflow(data = results_for_meta_analysis_h2,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0)influence_meta_results_h2 <- influence(meta_results_h2$fitted_model)
print(influence_meta_results_h2, digits = 2) ##
## rstudent dffits cook.d cov.r tau2.del QE.del hat
## Experiment 1 1.20 0.41 0.16 1.07 0.28 62.11 0.11
## Experiment 2 -3.60 -1.39 0.75 0.43 0.09 22.72 0.11
## Experiment 3 0.23 0.09 0.01 1.27 0.34 68.34 0.11
## Experiment 4 0.45 0.17 0.03 1.27 0.33 67.28 0.12
## Experiment 5 0.28 0.11 0.01 1.29 0.34 68.20 0.12
## Experiment 6 1.13 0.41 0.16 1.10 0.28 55.98 0.12
## Experiment 7 0.29 0.11 0.01 1.30 0.34 67.71 0.12
## Experiment 8 (standard) -0.78 -0.26 0.07 1.17 0.31 63.26 0.10
## Experiment 8 (random) -0.66 -0.22 0.05 1.19 0.32 64.94 0.10
## weight dfbs inf
## Experiment 1 10.65 0.41
## Experiment 2 11.07 -1.40 *
## Experiment 3 10.78 0.08
## Experiment 4 11.56 0.17
## Experiment 5 11.54 0.11
## Experiment 6 11.67 0.41
## Experiment 7 12.18 0.11
## Experiment 8 (standard) 10.43 -0.26
## Experiment 8 (random) 10.12 -0.22
Experiment 2 was assessed as being an outlier.
A sensitivity analysis excluded Experiment 2 and re meta analysed.
results_for_meta_analysis_h2_sensitivity <- results_for_meta_analysis_h2 %>%
filter(Experiment != "Experiment 2")
meta_results_h2_sensitivity <-
meta_analysis_workflow(data = results_for_meta_analysis_h2_sensitivity,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0)results_for_meta_analysis_h3 <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3", "Experiment 4",
"Experiment 5", "Experiment 6", "Experiment 7"),
logOR = c(results_exp_1_h3$logOR[results_exp_1_h3$var == "Struan"],
results_exp_2_h3$logOR[results_exp_2_h3$var == "Struan"],
results_exp_3_h3$logOR[results_exp_3_h3$var == "Struan"],
results_exp_4_h3$logOR[results_exp_4_h3$var == "Struan"],
results_exp_5_h3$logOR[results_exp_5_h3$var == "Struan"],
results_exp_6_h3$logOR[results_exp_6_h3$var == "Struan"],
results_exp_7_h3$logOR[results_exp_7_h3$var == "Struan"]),
logOR_se = c(results_exp_1_h3$logOR_se[results_exp_1_h3$var == "Struan"],
results_exp_2_h3$logOR_se[results_exp_2_h3$var == "Struan"],
results_exp_3_h3$logOR_se[results_exp_3_h3$var == "Struan"],
results_exp_4_h3$logOR_se[results_exp_4_h3$var == "Struan"],
results_exp_5_h3$logOR_se[results_exp_5_h3$var == "Struan"],
results_exp_6_h3$logOR_se[results_exp_6_h3$var == "Struan"],
results_exp_7_h3$logOR_se[results_exp_7_h3$var == "Struan"])) %>%
mutate(yi = logOR,
sei = logOR_se)
results_for_meta_analysis_h3 %>%
select(Experiment, yi, sei) %>%
write_csv("../meta analysis data/results_for_meta_analysis_behavioural_intentions.csv")
meta_results_h3 <- meta_analysis_workflow(data = results_for_meta_analysis_h3,
effect_size_label = "Odds Ratio",
exp_estimates = TRUE,
reference_line = 0)influence_meta_results_h3 <- influence(meta_results_h3$fitted_model)
print(influence_meta_results_h3, digits = 2) ##
## rstudent dffits cook.d cov.r tau2.del QE.del hat weight dfbs inf
## Experiment 1 0.75 0.29 0.09 1.24 1.83 39.97 0.13 12.83 0.29
## Experiment 2 -4.78 -1.67 0.67 0.22 0.16 9.31 0.14 13.81 -1.86 *
## Experiment 3 0.27 0.11 0.01 1.37 2.03 41.95 0.13 13.31 0.11
## Experiment 4 0.35 0.14 0.02 1.39 2.03 41.27 0.15 14.61 0.14
## Experiment 5 0.04 0.02 0.00 1.42 2.09 42.25 0.14 14.38 0.02
## Experiment 6 0.90 0.38 0.15 1.22 1.74 32.68 0.15 15.10 0.38
## Experiment 7 -0.06 -0.03 0.00 1.46 2.12 41.59 0.16 15.95 -0.03
Experiment 2 was assessed as being an outlier.
A sensitivity analysis excluded Experiment 2 and re meta analysed.
results_for_meta_analysis_h3_sensitivity <- results_for_meta_analysis_h3 %>%
filter(Experiment != "Experiment 2")
meta_results_h3_sensitivity <-
meta_analysis_workflow(data = results_for_meta_analysis_h3_sensitivity,
effect_size_label = "Odds Ratio",
exp_estimates = TRUE,
reference_line = 0)Refit meta analyses after excluding subsets of participants based on the exploratory awareness variables.
Note that these robustness tests also take the influence tests into account by excluding Experiment 2.
# re run tests with desired subsets of participants
## exp 1
results_exp_1_h1_us_contingency_memory <- data_exp_1 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h1_color_valence_contingency_memory <- data_exp_1 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h1_demand_compliance <- data_exp_1 %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h1_reactance_IAT <- data_exp_1 %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 2
results_exp_2_h1_us_contingency_memory <- data_exp_2 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h1_color_valence_contingency_memory <- data_exp_2 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h1_demand_compliance <- data_exp_2 %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h1_reactance_IAT <- data_exp_2 %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 3
results_exp_3_h1_us_contingency_memory <- data_exp_3 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h1_color_valence_contingency_memory <- data_exp_3 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h1_demand_compliance <- data_exp_3 %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h1_reactance_IAT <- data_exp_3 %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 4
results_exp_4_h1_us_contingency_memory <- data_exp_4 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h1_color_valence_contingency_memory <- data_exp_4 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h1_demand_compliance <- data_exp_4 %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h1_reactance_IAT <- data_exp_4 %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 5
results_exp_5_h1_us_contingency_memory <- data_exp_5 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h1_color_valence_contingency_memory <- data_exp_5 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h1_demand_compliance <- data_exp_5 %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h1_reactance_IAT <- data_exp_5 %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 6
results_exp_6_h1_demand_compliance <- data_exp_6 %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_6_h1_reactance_IAT <- data_exp_6 %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 standard instructions
results_exp_8nri_h1_us_contingency_memory <- data_exp_8nri %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8nri_h1_demand_compliance <- data_exp_8nri %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8nri_h1_reactance_IAT <- data_exp_8nri %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 random instructions
results_exp_8ri_h1_us_contingency_memory <- data_exp_8ri %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8ri_h1_demand_compliance <- data_exp_8ri %>%
filter(demand_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8ri_h1_reactance_IAT <- data_exp_8ri %>%
filter(reactance_IAT == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
# US contingency_memory
results_for_meta_analysis_h1_us_contingency_memory <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_2_h1_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_3_h1_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_4_h1_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_5_h1_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_8nri_h1_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_8ri_h1_us_contingency_memory$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_2_h1_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_3_h1_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_4_h1_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_5_h1_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_8nri_h1_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_8ri_h1_us_contingency_memory$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_2_h1_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_3_h1_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_4_h1_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_5_h1_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_8nri_h1_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_8ri_h1_us_contingency_memory$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# color valence contingency_memory
results_for_meta_analysis_h1_color_valence_contingency_memory <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5"),
cohens_d = c(results_exp_1_h1_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_2_h1_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_3_h1_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_4_h1_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_5_h1_color_valence_contingency_memory$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_2_h1_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_3_h1_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_4_h1_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_5_h1_color_valence_contingency_memory$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_2_h1_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_3_h1_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_4_h1_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_5_h1_color_valence_contingency_memory$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# demand_compliance
results_for_meta_analysis_h1_demand_compliance <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_2_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_3_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_4_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_5_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_6_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_8nri_h1_demand_compliance$cohens_d$estimate[[1]],
results_exp_8ri_h1_demand_compliance$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_2_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_3_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_4_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_5_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_6_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_8nri_h1_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_8ri_h1_demand_compliance$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_2_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_3_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_4_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_5_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_6_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_8nri_h1_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_8ri_h1_demand_compliance$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# reactance
results_for_meta_analysis_h1_reactance_IAT <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_2_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_3_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_4_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_5_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_6_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_8nri_h1_reactance_IAT$cohens_d$estimate[[1]],
results_exp_8ri_h1_reactance_IAT$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_2_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_3_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_4_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_5_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_6_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_8nri_h1_reactance_IAT$cohens_d$conf.int[["lower"]],
results_exp_8ri_h1_reactance_IAT$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_2_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_3_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_4_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_5_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_6_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_8nri_h1_reactance_IAT$cohens_d$conf.int[["upper"]],
results_exp_8ri_h1_reactance_IAT$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# fit metas
meta_results_h1_us_contingency_memory <-
meta_analysis_workflow(data = results_for_meta_analysis_h1_us_contingency_memory,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
meta_results_h1_color_valence_contingency_memory <-
meta_analysis_workflow(data = results_for_meta_analysis_h1_color_valence_contingency_memory,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
meta_results_h1_demand_compliance <-
meta_analysis_workflow(data = results_for_meta_analysis_h1_demand_compliance,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
meta_results_h1_reactance_IAT <-
meta_analysis_workflow(data = results_for_meta_analysis_h1_reactance_IAT,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
# robustness strings
robustness_us_contingency_memory_h1 <-
ifelse(robust_test(meta_results_h1_us_contingency_memory$meta_analysis_results,
meta_results_h1$meta_analysis_results),
"were", "were not")
robustness_color_valence_contingency_memory_h1 <-
ifelse(robust_test(meta_results_h1_color_valence_contingency_memory$meta_analysis_results,
meta_results_h1$meta_analysis_results),
"were", "were not")
robustness_demand_compliance_h1 <-
ifelse(robust_test(meta_results_h1_demand_compliance$meta_analysis_results,
meta_results_h1$meta_analysis_results),
"were", "were not")
robustness_reactance_IAT_excluded_h1 <-
ifelse(robust_test(meta_results_h1_reactance_IAT$meta_analysis_results,
meta_results_h1$meta_analysis_results),
"were", "were not")# re run tests with desired subsets of participants
## exp 1
results_exp_1_h2_us_contingency_memory <- data_exp_1 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h2_color_valence_contingency_memory <- data_exp_1 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h2_demand_compliance <- data_exp_1 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h2_reactance_self_reports <- data_exp_1 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 2
results_exp_2_h2_us_contingency_memory <- data_exp_2 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h2_color_valence_contingency_memory <- data_exp_2 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h2_demand_compliance <- data_exp_2 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h2_reactance_self_reports <- data_exp_2 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 3
results_exp_3_h2_us_contingency_memory <- data_exp_3 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h2_color_valence_contingency_memory <- data_exp_3 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h2_demand_compliance <- data_exp_3 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h2_reactance_self_reports <- data_exp_3 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 4
results_exp_4_h2_us_contingency_memory <- data_exp_4 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h2_color_valence_contingency_memory <- data_exp_4 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h2_demand_compliance <- data_exp_4 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h2_reactance_self_reports <- data_exp_4 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 5
results_exp_5_h2_us_contingency_memory <- data_exp_5 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h2_color_valence_contingency_memory <- data_exp_5 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h2_demand_compliance <- data_exp_5 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h2_reactance_self_reports <- data_exp_5 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 6
results_exp_6_h2_demand_compliance <- data_exp_6 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_6_h2_reactance_self_reports <- data_exp_6 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 7
results_exp_7_h2_us_contingency_memory <- data_exp_7 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_7_h2_color_valence_contingency_memory <- data_exp_7 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_7_h2_demand_compliance <- data_exp_7 %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_7_h2_reactance_self_reports <- data_exp_7 %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 standard instructions
results_exp_8nri_h2_us_contingency_memory <- data_exp_8nri %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8nri_h2_demand_compliance <- data_exp_8nri %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8nri_h2_reactance_self_reports <- data_exp_8nri %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 random instructions
results_exp_8ri_h2_us_contingency_memory <- data_exp_8ri %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8ri_h2_demand_compliance <- data_exp_8ri %>%
filter(demand_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8ri_h2_reactance_self_reports <- data_exp_8ri %>%
filter(reactance_self_reports == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
# US contingency_memory
results_for_meta_analysis_h2_us_contingency_memory <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5",
"Experiment 7",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_2_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_3_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_4_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_5_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_7_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_8nri_h2_us_contingency_memory$cohens_d$estimate[[1]],
results_exp_8ri_h2_us_contingency_memory$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_2_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_3_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_4_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_5_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_7_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_8nri_h2_us_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2_us_contingency_memory$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_2_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_3_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_4_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_5_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_7_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_8nri_h2_us_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2_us_contingency_memory$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# color valence contingency_memory
results_for_meta_analysis_h2_color_valence_contingency_memory <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5",
"Experiment 7"),
cohens_d = c(results_exp_1_h2_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_2_h2_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_3_h2_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_4_h2_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_5_h2_color_valence_contingency_memory$cohens_d$estimate[[1]],
results_exp_7_h2_color_valence_contingency_memory$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_2_h2_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_3_h2_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_4_h2_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_5_h2_color_valence_contingency_memory$cohens_d$conf.int[["lower"]],
results_exp_7_h2_color_valence_contingency_memory$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_2_h2_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_3_h2_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_4_h2_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_5_h2_color_valence_contingency_memory$cohens_d$conf.int[["upper"]],
results_exp_7_h2_color_valence_contingency_memory$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# demand_compliance
results_for_meta_analysis_h2_demand_compliance <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 7",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_2_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_3_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_4_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_5_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_6_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_7_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_8nri_h2_demand_compliance$cohens_d$estimate[[1]],
results_exp_8ri_h2_demand_compliance$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_2_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_3_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_4_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_5_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_6_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_7_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_8nri_h2_demand_compliance$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2_demand_compliance$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_2_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_3_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_4_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_5_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_6_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_7_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_8nri_h2_demand_compliance$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2_demand_compliance$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# reactance_self_reports
results_for_meta_analysis_h2_reactance_self_reports <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 7",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_2_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_3_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_4_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_5_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_6_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_7_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_8nri_h2_reactance_self_reports$cohens_d$estimate[[1]],
results_exp_8ri_h2_reactance_self_reports$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_2_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_3_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_4_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_5_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_6_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_7_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_8nri_h2_reactance_self_reports$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2_reactance_self_reports$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_2_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_3_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_4_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_5_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_6_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_7_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_8nri_h2_reactance_self_reports$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2_reactance_self_reports$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# fit metas
meta_results_h2_us_contingency_memory <-
meta_analysis_workflow(data = results_for_meta_analysis_h2_us_contingency_memory,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
meta_results_h2_color_valence_contingency_memory <-
meta_analysis_workflow(data = results_for_meta_analysis_h2_color_valence_contingency_memory,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
meta_results_h2_demand_compliance <-
meta_analysis_workflow(data = results_for_meta_analysis_h2_demand_compliance,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
meta_results_h2_reactance_self_reports <-
meta_analysis_workflow(data = results_for_meta_analysis_h2_reactance_self_reports,
effect_size_label = "Cohen's d",
exp_estimates = FALSE,
reference_line = 0,
plot = FALSE)
# robustness strings
robustness_us_contingency_memory_h2 <-
ifelse(robust_test(meta_results_h2_us_contingency_memory$meta_analysis_results,
meta_results_h2$meta_analysis_results),
"were", "were not")
robustness_color_valence_contingency_memory_h2 <-
ifelse(robust_test(meta_results_h2_color_valence_contingency_memory$meta_analysis_results,
meta_results_h2$meta_analysis_results),
"were", "were not")
robustness_demand_compliance_h2 <-
ifelse(robust_test(meta_results_h2_demand_compliance$meta_analysis_results,
meta_results_h2$meta_analysis_results),
"were", "were not")
robustness_reactance_self_reports_h2 <-
ifelse(robust_test(meta_results_h2_reactance_self_reports$meta_analysis_results,
meta_results_h2$meta_analysis_results),
"were", "were not")# re run tests with desired subsets of participants
# exp 1
results_exp_1_h3_us_contingency_memory <- data_exp_1 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_1_h3_color_valence_contingency_memory <- data_exp_1 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 2
results_exp_2_h3_us_contingency_memory <- data_exp_2 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_2_h3_color_valence_contingency_memory <- data_exp_2 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 3
results_exp_3_h3_us_contingency_memory <- data_exp_3 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_3_h3_color_valence_contingency_memory <- data_exp_3 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 4
results_exp_4_h3_us_contingency_memory <- data_exp_4 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_4_h3_color_valence_contingency_memory <- data_exp_4 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 5
results_exp_5_h3_us_contingency_memory <- data_exp_5 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_5_h3_color_valence_contingency_memory <- data_exp_5 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 7
results_exp_7_h3_us_contingency_memory <- data_exp_7 %>%
filter(US_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_7_h3_color_valence_contingency_memory <- data_exp_7 %>%
filter(color_valence_contingency_memory == TRUE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# us contingency memory
results_for_meta_analysis_h3_us_contingency_memory <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3", "Experiment 4",
"Experiment 5", "Experiment 7"),
logOR = c(results_exp_1_h3_us_contingency_memory$logOR[results_exp_1_h3_us_contingency_memory$var == "Struan"],
results_exp_2_h3_us_contingency_memory$logOR[results_exp_2_h3_us_contingency_memory$var == "Struan"],
results_exp_3_h3_us_contingency_memory$logOR[results_exp_3_h3_us_contingency_memory$var == "Struan"],
results_exp_4_h3_us_contingency_memory$logOR[results_exp_4_h3_us_contingency_memory$var == "Struan"],
results_exp_5_h3_us_contingency_memory$logOR[results_exp_5_h3_us_contingency_memory$var == "Struan"],
results_exp_7_h3_us_contingency_memory$logOR[results_exp_7_h3_us_contingency_memory$var == "Struan"]),
logOR_se = c(results_exp_1_h3_us_contingency_memory$logOR_se[results_exp_1_h3_us_contingency_memory$var == "Struan"],
results_exp_2_h3_us_contingency_memory$logOR_se[results_exp_2_h3_us_contingency_memory$var == "Struan"],
results_exp_3_h3_us_contingency_memory$logOR_se[results_exp_3_h3_us_contingency_memory$var == "Struan"],
results_exp_4_h3_us_contingency_memory$logOR_se[results_exp_4_h3_us_contingency_memory$var == "Struan"],
results_exp_5_h3_us_contingency_memory$logOR_se[results_exp_5_h3_us_contingency_memory$var == "Struan"],
results_exp_7_h3_us_contingency_memory$logOR_se[results_exp_7_h3_us_contingency_memory$var == "Struan"])) %>%
mutate(yi = logOR,
sei = logOR_se) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# color valence contingency memory
results_for_meta_analysis_h3_color_valence_contingency_memory <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3", "Experiment 4",
"Experiment 5", "Experiment 7"),
logOR = c(results_exp_1_h3_color_valence_contingency_memory$logOR[results_exp_1_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_2_h3_color_valence_contingency_memory$logOR[results_exp_2_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_3_h3_color_valence_contingency_memory$logOR[results_exp_3_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_4_h3_color_valence_contingency_memory$logOR[results_exp_4_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_5_h3_color_valence_contingency_memory$logOR[results_exp_5_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_7_h3_color_valence_contingency_memory$logOR[results_exp_7_h3_color_valence_contingency_memory$var == "Struan"]),
logOR_se = c(results_exp_1_h3_color_valence_contingency_memory$logOR_se[results_exp_1_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_2_h3_color_valence_contingency_memory$logOR_se[results_exp_2_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_3_h3_color_valence_contingency_memory$logOR_se[results_exp_3_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_4_h3_color_valence_contingency_memory$logOR_se[results_exp_4_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_5_h3_color_valence_contingency_memory$logOR_se[results_exp_5_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_7_h3_color_valence_contingency_memory$logOR_se[results_exp_7_h3_color_valence_contingency_memory$var == "Struan"])) %>%
mutate(yi = logOR,
sei = logOR_se) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2")
# meta fits
meta_results_h3_us_contingency_memory <-
meta_analysis_workflow(data = results_for_meta_analysis_h3_us_contingency_memory,
effect_size_label = "Odds Ratio",
exp_estimates = TRUE,
reference_line = 0,
plot = FALSE)
meta_results_h3_color_valence_contingency_memory <-
meta_analysis_workflow(data = results_for_meta_analysis_h3_color_valence_contingency_memory,
effect_size_label = "Odds Ratio",
exp_estimates = TRUE,
reference_line = 0,
plot = FALSE)
# robustness strings
robustness_us_contingency_memory_h3 <-
ifelse(robust_test(meta_results_h3_us_contingency_memory$meta_analysis_results,
meta_results_h1$meta_analysis_results),
"were", "were not")
robustness_color_valence_contingency_memory_h3 <-
ifelse(robust_test(meta_results_h3_color_valence_contingency_memory$meta_analysis_results,
meta_results_h1$meta_analysis_results),
"were", "were not")Our robustness analyses exclude participants who meet certain criteria. A reviewer requested that we also calculate modration by these variables.
Note that these tests also take the influence tests into account by excluding Experiment 2.
# re run tests with desired subsets of participants
## exp 1
results_exp_1_h1_us_contingency_memory_excluded <- data_exp_1 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h1_color_valence_contingency_memory_excluded <- data_exp_1 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h1_demand_compliance_excluded <- data_exp_1 %>%
filter(demand_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h1_reactance_IAT_excluded <- data_exp_1 %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 2
results_exp_2_h1_us_contingency_memory_excluded <- data_exp_2 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h1_color_valence_contingency_memory_excluded <- data_exp_2 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h1_demand_compliance_excluded <- data_exp_2 %>%
filter(demand_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h1_reactance_IAT_excluded <- data_exp_2 %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 3
results_exp_3_h1_us_contingency_memory_excluded <- data_exp_3 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h1_color_valence_contingency_memory_excluded <- data_exp_3 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h1_demand_compliance_excluded <- data_exp_3 %>%
filter(demand_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h1_reactance_IAT_excluded <- data_exp_3 %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 4
results_exp_4_h1_us_contingency_memory_excluded <- data_exp_4 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h1_color_valence_contingency_memory_excluded <- data_exp_4 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h1_demand_compliance_excluded <- data_exp_4 %>%
filter(demand_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h1_reactance_IAT_excluded <- data_exp_4 %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 5
results_exp_5_h1_us_contingency_memory_excluded <- data_exp_5 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h1_color_valence_contingency_memory_excluded <- data_exp_5 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h1_demand_compliance_excluded <- data_exp_5 %>%
filter(demand_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h1_reactance_IAT_excluded <- data_exp_5 %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
# ## exp 6
# results_exp_6_h1_demand_compliance_excluded <- data_exp_6 %>%
# filter(demand_IAT == TRUE) %>%
# mutate(DV = IAT_Score,
# IV = source_valence_condition) %>%
# t_test_analyses(.)
#>not enough 'x' observations
results_exp_6_h1_reactance_IAT_excluded <- data_exp_6 %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 standard instructions
# results_exp_8nri_h1_us_contingency_memory_excluded <- data_exp_8nri %>%
# filter(US_contingency_memory == FALSE) %>%
# mutate(DV = IAT_Score,
# IV = source_valence_condition) %>%
# t_test_analyses(.)
#>not enough 'x' observations
# results_exp_8nri_h1_demand_compliance_excluded <- data_exp_8nri %>%
# filter(demand_IAT == TRUE) %>%
# mutate(DV = IAT_Score,
# IV = source_valence_condition) %>%
# t_test_analyses(.)
#>not enough 'x' observations
results_exp_8nri_h1_reactance_IAT_excluded <- data_exp_8nri %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 random instructions
results_exp_8ri_h1_us_contingency_memory_excluded <- data_exp_8ri %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
# results_exp_8ri_h1_demand_compliance_excluded <- data_exp_8ri %>%
# filter(demand_IAT == TRUE) %>%
# distinct(source_valence_condition)
# mutate(DV = IAT_Score,
# IV = source_valence_condition) %>%
# t_test_analyses(.)
# not enough 'x' observations
results_exp_8ri_h1_reactance_IAT_excluded <- data_exp_8ri %>%
filter(reactance_IAT == TRUE) %>%
mutate(DV = IAT_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
# US contingency_memory
results_for_meta_analysis_h1_us_contingency_memory_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5",
#"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_2_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_3_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_4_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_5_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]],
#results_exp_8nri_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_8ri_h1_us_contingency_memory_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
#results_exp_8nri_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_8ri_h1_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
#results_exp_8nri_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_8ri_h1_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# color valence contingency_memory
results_for_meta_analysis_h1_color_valence_contingency_memory_excluded <-
data.frame(Experiment = c("Experiment 1",
"Experiment 2",
"Experiment 3",
"Experiment 4",
"Experiment 5"),
cohens_d = c(results_exp_1_h1_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_2_h1_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_3_h1_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_4_h1_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_5_h1_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h1_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# demand_compliance
results_for_meta_analysis_h1_demand_compliance_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5"),
#"Experiment 6",
#"Experiment 8 (standard)"),
#"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_2_h1_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_3_h1_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_4_h1_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_5_h1_demand_compliance_excluded$cohens_d$estimate[[1]]),
#results_exp_6_h1_demand_compliance_excluded_excluded$cohens_d$estimate[[1]],
#results_exp_8nri_h1_demand_compliance_excluded$cohens_d$estimate[[1]]),
#results_exp_8ri_h1_demand_compliance_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]]),
#results_exp_6_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
#results_exp_8nri_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]]),
#results_exp_8ri_h1_demand_compliance_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h1_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h1_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h1_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h1_demand_compliance_excluded$cohens_d$conf.int[["upper"]])) %>%
#results_exp_6_h1_demand_compliance_excluded$cohens_d_excluded$conf.int[["upper"]],
#results_exp_8nri_h1_demand_compliance_excluded$cohens_d$conf.int[["upper"]])) %>%
#results_exp_8ri_h1_demand_compliance_excluded$cohens_d_excluded$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# reactance
results_for_meta_analysis_h1_reactance_IAT_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_2_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_3_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_4_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_5_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_6_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_8nri_h1_reactance_IAT_excluded$cohens_d$estimate[[1]],
results_exp_8ri_h1_reactance_IAT_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_6_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_8nri_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]],
results_exp_8ri_h1_reactance_IAT_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_6_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_8nri_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]],
results_exp_8ri_h1_reactance_IAT_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# fit metas
mod_meta_results_h1_us_contingency_memory <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h1_us_contingency_memory, exclude = FALSE),
results_for_meta_analysis_h1_us_contingency_memory_excluded,
by = "Experiment"),
results_for_meta_analysis_h1_us_contingency_memory_excluded),
effect_size_label = "Cohen's d")
mod_meta_results_h1_color_valence_contingency_memory <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h1_color_valence_contingency_memory, exclude = FALSE),
results_for_meta_analysis_h1_color_valence_contingency_memory_excluded,
by = "Experiment"),
results_for_meta_analysis_h1_color_valence_contingency_memory_excluded),
effect_size_label = "Cohen's d")
mod_meta_results_h1_demand_compliance <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h1_demand_compliance, exclude = FALSE),
results_for_meta_analysis_h1_demand_compliance_excluded,
by = "Experiment"),
results_for_meta_analysis_h1_demand_compliance_excluded),
effect_size_label = "Cohen's d")
mod_meta_results_h1_reactance <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h1_reactance_IAT, exclude = FALSE),
results_for_meta_analysis_h1_reactance_IAT_excluded,
by = "Experiment"),
results_for_meta_analysis_h1_reactance_IAT_excluded),
effect_size_label = "Cohen's d")
# forest plots
forest(mod_meta_results_h1_us_contingency_memory$fitted_model)Moderation by:
# re run tests with desired subsets of participants
## exp 1
results_exp_1_h2_us_contingency_memory_excluded <- data_exp_1 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h2_color_valence_contingency_memory_excluded <- data_exp_1 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h2_demand_compliance_excluded <- data_exp_1 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_1_h2_reactance_excluded <- data_exp_1 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 2
results_exp_2_h2_us_contingency_memory_excluded <- data_exp_2 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h2_color_valence_contingency_memory_excluded <- data_exp_2 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h2_demand_compliance_excluded <- data_exp_2 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_2_h2_reactance_self_reports_excluded <- data_exp_2 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 3
results_exp_3_h2_us_contingency_memory_excluded <- data_exp_3 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h2_color_valence_contingency_memory_excluded <- data_exp_3 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h2_demand_compliance_excluded <- data_exp_3 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_3_h2_reactance_self_reports_excluded <- data_exp_3 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 4
results_exp_4_h2_us_contingency_memory_excluded <- data_exp_4 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h2_color_valence_contingency_memory_excluded <- data_exp_4 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h2_demand_compliance_excluded <- data_exp_4 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_4_h2_reactance_self_reports_excluded <- data_exp_4 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 5
results_exp_5_h2_us_contingency_memory_excluded <- data_exp_5 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h2_color_valence_contingency_memory_excluded <- data_exp_5 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h2_demand_compliance_excluded <- data_exp_5 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_5_h2_reactance_self_reports_excluded <- data_exp_5 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 6
results_exp_6_h2_demand_compliance_excluded <- data_exp_6 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_6_h2_reactance_self_reports_excluded <- data_exp_6 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 7
results_exp_7_h2_us_contingency_memory_excluded <- data_exp_7 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_7_h2_color_valence_contingency_memory_excluded <- data_exp_7 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_7_h2_demand_compliance_excluded <- data_exp_7 %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_7_h2_reactance_self_reports_excluded <- data_exp_7 %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 standard instructions
# results_exp_8nri_h2_us_contingency_memory_excluded <- data_exp_8nri %>%
# filter(US_contingency_memory == FALSE) %>%
# mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
# IV = source_valence_condition) %>%
# t_test_analyses(.)
# > NOT ENOUGH OBSERVATIONS
results_exp_8nri_h2_demand_compliance_excluded <- data_exp_8nri %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8nri_h2_reactance_self_reports_excluded <- data_exp_8nri %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
## exp 8 random instructions
results_exp_8ri_h2_us_contingency_memory_excluded <- data_exp_8ri %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8ri_h2_demand_compliance_excluded <- data_exp_8ri %>%
filter(demand_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
results_exp_8ri_h2_reactance_self_reports_excluded <- data_exp_8ri %>%
filter(reactance_self_reports == TRUE) %>%
mutate(DV = EC_Effect_CS1_CS2_Difference_Score,
IV = source_valence_condition) %>%
t_test_analyses(.)
# US contingency_memory
results_for_meta_analysis_h2_us_contingency_memory_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5",
"Experiment 7",
#"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_2_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_3_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_4_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_5_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_7_h2_us_contingency_memory$cohens_d$estimate[[1]],
#results_exp_8nri_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_8ri_h2_us_contingency_memory_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_7_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
#results_exp_8nri_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2_us_contingency_memory_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_7_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
#results_exp_8nri_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2_us_contingency_memory_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# color valence contingency_memory
results_for_meta_analysis_h2_color_valence_contingency_memory_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5",
"Experiment 7"),
cohens_d = c(results_exp_1_h2_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_2_h2_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_3_h2_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_4_h2_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_5_h2_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]],
results_exp_7_h2_color_valence_contingency_memory_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]],
results_exp_7_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]],
results_exp_7_h2_color_valence_contingency_memory_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# demand_compliance
results_for_meta_analysis_h2_demand_compliance_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 7",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_2_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_3_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_4_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_5_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_6_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_7_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_8nri_h2_demand_compliance_excluded$cohens_d$estimate[[1]],
results_exp_8ri_h2_demand_compliance_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_6_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_7_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_8nri_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2_demand_compliance_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_6_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_7_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_8nri_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2_demand_compliance_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# reactance_self_reports
results_for_meta_analysis_h2_reactance_self_reports_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3",
"Experiment 4", "Experiment 5", "Experiment 6",
"Experiment 7",
"Experiment 8 (standard)",
"Experiment 8 (random)"),
cohens_d = c(results_exp_1_h2_reactance_excluded$cohens_d$estimate[[1]],
results_exp_2_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_3_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_4_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_5_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_6_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_7_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_8nri_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]],
results_exp_8ri_h2_reactance_self_reports_excluded$cohens_d$estimate[[1]]),
ci_lwr = c(results_exp_1_h2_reactance_excluded$cohens_d$conf.int[["lower"]],
results_exp_2_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_3_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_4_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_5_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_6_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_7_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_8nri_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]],
results_exp_8ri_h2_reactance_self_reports_excluded$cohens_d$conf.int[["lower"]]),
ci_upr = c(results_exp_1_h2_reactance_excluded$cohens_d$conf.int[["upper"]],
results_exp_2_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_3_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_4_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_5_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_6_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_7_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_8nri_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]],
results_exp_8ri_h2_reactance_self_reports_excluded$cohens_d$conf.int[["upper"]])) %>%
mutate(yi = cohens_d,
sei = (ci_upr - ci_lwr)/(1.96*2)) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# fit metas
mod_meta_results_h2_us_contingency_memory <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h2_us_contingency_memory, exclude = FALSE),
results_for_meta_analysis_h2_us_contingency_memory_excluded,
by = "Experiment"),
results_for_meta_analysis_h2_us_contingency_memory_excluded),
effect_size_label = "Cohen's d")
mod_meta_results_h2_color_valence_contingency_memory <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h2_color_valence_contingency_memory, exclude = FALSE),
results_for_meta_analysis_h2_color_valence_contingency_memory_excluded,
by = "Experiment"),
results_for_meta_analysis_h2_color_valence_contingency_memory_excluded),
effect_size_label = "Cohen's d")
mod_meta_results_h2_demand_compliance <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h2_demand_compliance, exclude = FALSE),
results_for_meta_analysis_h2_demand_compliance_excluded,
by = "Experiment"),
results_for_meta_analysis_h2_demand_compliance_excluded),
effect_size_label = "Cohen's d")
mod_meta_results_h2_reactance <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h2_reactance_self_reports, exclude = FALSE),
results_for_meta_analysis_h2_reactance_self_reports_excluded,
by = "Experiment"),
results_for_meta_analysis_h2_reactance_self_reports_excluded),
effect_size_label = "Cohen's d")
# forest plots
forest(mod_meta_results_h2_us_contingency_memory$fitted_model)Moderation by:
# re run tests with desired subsets of participants
# exp 1
results_exp_1_h3_us_contingency_memory_excluded <- data_exp_1 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_1_h3_color_valence_contingency_memory_excluded <- data_exp_1 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 2
results_exp_2_h3_us_contingency_memory_excluded <- data_exp_2 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_2_h3_color_valence_contingency_memory_excluded <- data_exp_2 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 3
results_exp_3_h3_us_contingency_memory_excluded <- data_exp_3 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_3_h3_color_valence_contingency_memory_excluded <- data_exp_3 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 4
results_exp_4_h3_us_contingency_memory_excluded <- data_exp_4 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_4_h3_color_valence_contingency_memory_excluded <- data_exp_4 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 5
results_exp_5_h3_us_contingency_memory_excluded <- data_exp_5 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_5_h3_color_valence_contingency_memory_excluded <- data_exp_5 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# exp 7
results_exp_7_h3_us_contingency_memory_excluded <- data_exp_7 %>%
filter(US_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
results_exp_7_h3_color_valence_contingency_memory_excluded <- data_exp_7 %>%
filter(color_valence_contingency_memory == FALSE) %>%
mutate(DV = intention,
IV = source_valence_condition) %>%
multinominal_analysis()
# us contingency memory
results_for_meta_analysis_h3_us_contingency_memory_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3", "Experiment 4",
"Experiment 5", "Experiment 7"),
logOR = c(results_exp_1_h3_us_contingency_memory_excluded$logOR[results_exp_1_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_2_h3_us_contingency_memory_excluded$logOR[results_exp_2_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_3_h3_us_contingency_memory_excluded$logOR[results_exp_3_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_4_h3_us_contingency_memory_excluded$logOR[results_exp_4_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_5_h3_us_contingency_memory_excluded$logOR[results_exp_5_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_7_h3_us_contingency_memory_excluded$logOR[results_exp_7_h3_us_contingency_memory_excluded$var == "Struan"]),
logOR_se = c(results_exp_1_h3_us_contingency_memory_excluded$logOR_se[results_exp_1_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_2_h3_us_contingency_memory_excluded$logOR_se[results_exp_2_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_3_h3_us_contingency_memory_excluded$logOR_se[results_exp_3_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_4_h3_us_contingency_memory_excluded$logOR_se[results_exp_4_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_5_h3_us_contingency_memory_excluded$logOR_se[results_exp_5_h3_us_contingency_memory_excluded$var == "Struan"],
results_exp_7_h3_us_contingency_memory_excluded$logOR_se[results_exp_7_h3_us_contingency_memory_excluded$var == "Struan"])) %>%
mutate(yi = logOR,
sei = logOR_se) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# color valence contingency memory
results_for_meta_analysis_h3_color_valence_contingency_memory_excluded <-
data.frame(Experiment = c("Experiment 1", "Experiment 2", "Experiment 3", "Experiment 4",
"Experiment 5", "Experiment 7"),
logOR = c(results_exp_1_h3_color_valence_contingency_memory_excluded$logOR[results_exp_1_h3_color_valence_contingency_memory_excluded$var == "Struan"],
results_exp_2_h3_color_valence_contingency_memory_excluded$logOR[results_exp_2_h3_color_valence_contingency_memory_excluded$var == "Struan"],
results_exp_3_h3_color_valence_contingency_memory_excluded$logOR[results_exp_3_h3_color_valence_contingency_memory_excluded$var == "Struan"],
results_exp_4_h3_color_valence_contingency_memory_excluded$logOR[results_exp_4_h3_color_valence_contingency_memory_excluded$var == "Struan"],
results_exp_5_h3_color_valence_contingency_memory_excluded$logOR[results_exp_5_h3_color_valence_contingency_memory_excluded$var == "Struan"],
results_exp_7_h3_color_valence_contingency_memory_excluded$logOR[results_exp_7_h3_color_valence_contingency_memory_excluded$var == "Struan"]),
logOR_se = c(results_exp_1_h3_color_valence_contingency_memory_excluded$logOR_se[results_exp_1_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_2_h3_color_valence_contingency_memory_excluded$logOR_se[results_exp_2_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_3_h3_color_valence_contingency_memory_excluded$logOR_se[results_exp_3_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_4_h3_color_valence_contingency_memory_excluded$logOR_se[results_exp_4_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_5_h3_color_valence_contingency_memory_excluded$logOR_se[results_exp_5_h3_color_valence_contingency_memory$var == "Struan"],
results_exp_7_h3_color_valence_contingency_memory_excluded$logOR_se[results_exp_7_h3_color_valence_contingency_memory$var == "Struan"])) %>%
mutate(yi = logOR,
sei = logOR_se) %>%
# remove exp 2 on the basis of the influence test
filter(Experiment != "Experiment 2") %>%
mutate(exclude = TRUE)
# meta fits
mod_meta_results_h3_us_contingency_memory <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h3_us_contingency_memory, exclude = FALSE),
results_for_meta_analysis_h3_us_contingency_memory_excluded,
by = "Experiment"),
results_for_meta_analysis_h3_us_contingency_memory_excluded),
effect_size_label = "Odds Ratio")
mod_meta_results_h3_color_valence_contingency_memory <-
multilevel_moderator_meta_analysis_workflow(data = bind_rows(semi_join(mutate(results_for_meta_analysis_h3_color_valence_contingency_memory, exclude = FALSE),
results_for_meta_analysis_h3_color_valence_contingency_memory_excluded,
by = "Experiment"),
results_for_meta_analysis_h3_color_valence_contingency_memory_excluded),
effect_size_label = "Odds Ratio")
# forest plots
forest(mod_meta_results_h3_us_contingency_memory$fitted_model)Moderation by:
effect_sizes_by_subgroup <-
bind_rows(
# IATs
(data.frame(moderator = c("us_contingency_memory AWARE",
"color_valence_contingency_memory AWARE",
"demand_NONcompliance",
"NONreactant"),
es_robustness = c(mod_meta_results_h1_us_contingency_memory$fitted_model$b[1],
mod_meta_results_h1_color_valence_contingency_memory$fitted_model$b[1],
mod_meta_results_h1_demand_compliance$fitted_model$b[1],
mod_meta_results_h1_reactance$fitted_model$b[1]),
ci_lower_robustness = c(mod_meta_results_h1_us_contingency_memory$fitted_model$ci.lb[1],
mod_meta_results_h1_color_valence_contingency_memory$fitted_model$ci.lb[1],
mod_meta_results_h1_demand_compliance$fitted_model$ci.lb[1],
mod_meta_results_h1_reactance$fitted_model$ci.lb[1]),
ci_upper_robustness = c(mod_meta_results_h1_us_contingency_memory$fitted_model$ci.ub[1],
mod_meta_results_h1_color_valence_contingency_memory$fitted_model$ci.ub[1],
mod_meta_results_h1_demand_compliance$fitted_model$ci.ub[1],
mod_meta_results_h1_reactance$fitted_model$ci.ub[1]),
es_excluded = c(sum(mod_meta_results_h1_us_contingency_memory$fitted_model$b),
sum(mod_meta_results_h1_color_valence_contingency_memory$fitted_model$b),
sum(mod_meta_results_h1_demand_compliance$fitted_model$b),
sum(mod_meta_results_h1_reactance$fitted_model$b)),
ci_lower_excluded = c(sum(mod_meta_results_h1_us_contingency_memory$fitted_model$ci.lb),
sum(mod_meta_results_h1_color_valence_contingency_memory$fitted_model$ci.lb),
sum(mod_meta_results_h1_demand_compliance$fitted_model$ci.lb),
sum(mod_meta_results_h1_reactance$fitted_model$ci.lb)),
ci_upper_excluded = c(sum(mod_meta_results_h1_us_contingency_memory$fitted_model$ci.ub),
sum(mod_meta_results_h1_color_valence_contingency_memory$fitted_model$ci.ub),
sum(mod_meta_results_h1_demand_compliance$fitted_model$ci.ub),
sum(mod_meta_results_h1_reactance$fitted_model$ci.ub))) %>%
mutate(DV = "IAT",
es = "d")),
# Self reported evlautions
(data.frame(moderator = c("us_contingency_memory AWARE",
"color_valence_contingency_memory AWARE",
"demand_NONcompliance",
"NONreactant"),
es_robustness = c(mod_meta_results_h2_us_contingency_memory$fitted_model$b[1],
mod_meta_results_h2_color_valence_contingency_memory$fitted_model$b[1],
mod_meta_results_h2_demand_compliance$fitted_model$b[1],
mod_meta_results_h2_reactance$fitted_model$b[1]),
ci_lower_robustness = c(mod_meta_results_h2_us_contingency_memory$fitted_model$ci.lb[1],
mod_meta_results_h2_color_valence_contingency_memory$fitted_model$ci.lb[1],
mod_meta_results_h2_demand_compliance$fitted_model$ci.lb[1],
mod_meta_results_h2_reactance$fitted_model$ci.lb[1]),
ci_upper_robustness = c(mod_meta_results_h2_us_contingency_memory$fitted_model$ci.ub[1],
mod_meta_results_h2_color_valence_contingency_memory$fitted_model$ci.ub[1],
mod_meta_results_h2_demand_compliance$fitted_model$ci.ub[1],
mod_meta_results_h2_reactance$fitted_model$ci.ub[1]),
es_excluded = c(sum(mod_meta_results_h2_us_contingency_memory$fitted_model$b),
sum(mod_meta_results_h2_color_valence_contingency_memory$fitted_model$b),
sum(mod_meta_results_h2_demand_compliance$fitted_model$b),
sum(mod_meta_results_h2_reactance$fitted_model$b)),
ci_lower_excluded = c(sum(mod_meta_results_h2_us_contingency_memory$fitted_model$ci.lb),
sum(mod_meta_results_h2_color_valence_contingency_memory$fitted_model$ci.lb),
sum(mod_meta_results_h2_demand_compliance$fitted_model$ci.lb),
sum(mod_meta_results_h2_reactance$fitted_model$ci.lb)),
ci_upper_excluded = c(sum(mod_meta_results_h2_us_contingency_memory$fitted_model$ci.ub),
sum(mod_meta_results_h2_color_valence_contingency_memory$fitted_model$ci.ub),
sum(mod_meta_results_h2_demand_compliance$fitted_model$ci.ub),
sum(mod_meta_results_h2_reactance$fitted_model$ci.ub))) %>%
mutate(DV = "self reported evaluations",
es = "d")),
# Behavioural intentions
(data.frame(moderator = c("us_contingency_memory AWARE",
"color_valence_contingency_memory AWARE"),
es_robustness = c(mod_meta_results_h3_us_contingency_memory$fitted_model$b[1],
mod_meta_results_h3_color_valence_contingency_memory$fitted_model$b[1]),
ci_lower_robustness = c(mod_meta_results_h3_us_contingency_memory$fitted_model$ci.lb[1],
mod_meta_results_h3_color_valence_contingency_memory$fitted_model$ci.lb[1]),
ci_upper_robustness = c(mod_meta_results_h3_us_contingency_memory$fitted_model$ci.ub[1],
mod_meta_results_h3_color_valence_contingency_memory$fitted_model$ci.ub[1]),
es_excluded = c(sum(mod_meta_results_h3_us_contingency_memory$fitted_model$b),
sum(mod_meta_results_h3_color_valence_contingency_memory$fitted_model$b)),
ci_lower_excluded = c(sum(mod_meta_results_h3_us_contingency_memory$fitted_model$ci.lb),
sum(mod_meta_results_h3_color_valence_contingency_memory$fitted_model$ci.lb)),
ci_upper_excluded = c(sum(mod_meta_results_h3_us_contingency_memory$fitted_model$ci.ub),
sum(mod_meta_results_h3_color_valence_contingency_memory$fitted_model$ci.ub))) %>%
mutate(DV = "behvioural intentions",
es = "OR"))
) %>%
round_df(2) %>%
select(DV,
moderator,
es,
es_robustness,
ci_lower_robustness,
ci_upper_robustness,
es_excluded,
ci_lower_excluded,
ci_upper_excluded)
effect_sizes_by_subgroup %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| DV | moderator | es | es_robustness | ci_lower_robustness | ci_upper_robustness | es_excluded | ci_lower_excluded | ci_upper_excluded |
|---|---|---|---|---|---|---|---|---|
| IAT | us_contingency_memory AWARE | d | 0.83 | 0.54 | 1.12 | 0.85 | 0.15 | 1.54 |
| IAT | color_valence_contingency_memory AWARE | d | 1.11 | 0.46 | 1.77 | 0.33 | -1.54 | 2.20 |
| IAT | demand_NONcompliance | d | 0.80 | 0.46 | 1.13 | 0.76 | -0.16 | 1.69 |
| IAT | NONreactant | d | 0.85 | 0.67 | 1.04 | 0.83 | 0.13 | 1.52 |
| self reported evaluations | us_contingency_memory AWARE | d | 1.19 | 1.02 | 1.37 | 1.29 | 0.80 | 1.77 |
| self reported evaluations | color_valence_contingency_memory AWARE | d | 1.80 | 1.21 | 2.40 | 0.34 | -1.43 | 2.11 |
| self reported evaluations | demand_NONcompliance | d | 1.05 | 0.79 | 1.30 | 1.80 | 1.18 | 2.43 |
| self reported evaluations | NONreactant | d | 1.19 | 0.94 | 1.44 | 0.92 | 0.33 | 1.51 |
| behvioural intentions | us_contingency_memory AWARE | OR | 1.89 | 1.47 | 2.32 | 0.96 | -0.25 | 2.18 |
| behvioural intentions | color_valence_contingency_memory AWARE | OR | 2.88 | 1.95 | 3.82 | 0.03 | -2.64 | 2.69 |
combined_data <- bind_rows(mutate(data_exp_1, experiment = "1"),
#mutate(data_exp_2, experiment = "2"), # excluded by robustness tests
mutate(data_exp_3, experiment = "3"),
mutate(data_exp_4, experiment = "4"),
mutate(data_exp_5, experiment = "5"),
mutate(data_exp_6, experiment = "6"),
mutate(data_exp_7, experiment = "7"),
mutate(data_exp_8ri, experiment = "8ri"),
mutate(data_exp_8nri, experiment = "8nri"))
# n per cells
n_per_cell <-
bind_rows(
# IATs
# US_contingency_memory
(combined_data %>%
# EXCLUDE CONDITIONS THAT HAD INADEQUATE N TO CALCULATE ES IN METAS ABOVE
filter((!experiment %in% c("8nri") & US_contingency_memory == TRUE) |
US_contingency_memory == FALSE) %>%
select(IAT_Score, source_valence_condition, US_contingency_memory) %>%
drop_na() %>%
count(US_contingency_memory) %>%
mutate(moderator = "US_contingency_memory",
DV = "IAT") %>%
spread(US_contingency_memory, n) %>%
rename(n_robustness_subset = `TRUE`,
n_excluded_subset = `FALSE`)),
# color_valence_contingency_memory
(combined_data %>%
select(IAT_Score, source_valence_condition, color_valence_contingency_memory) %>%
drop_na() %>%
count(color_valence_contingency_memory) %>%
mutate(moderator = "color_valence_contingency_memory",
DV = "IAT") %>%
spread(color_valence_contingency_memory, n) %>%
rename(n_robustness_subset = `TRUE`,
n_excluded_subset = `FALSE`)),
# demand_IAT
(combined_data %>%
# EXCLUDE CONDITIONS THAT HAD INADEQUATE N TO CALCULATE ES IN METAS ABOVE
filter((!experiment %in% c("6", "8ri", "8nri") & demand_IAT == TRUE) |
demand_IAT == FALSE) %>%
select(IAT_Score, source_valence_condition, demand_IAT) %>%
drop_na() %>%
count(demand_IAT) %>%
mutate(moderator = "demand",
DV = "IAT") %>%
spread(demand_IAT, n) %>%
rename(n_robustness_subset = `FALSE`,
n_excluded_subset = `TRUE`)),
# reactance_IAT
(combined_data %>%
select(IAT_Score, source_valence_condition, reactance_IAT) %>%
drop_na() %>%
count(reactance_IAT) %>%
mutate(moderator = "reactance",
DV = "IAT") %>%
spread(reactance_IAT, n) %>%
rename(n_robustness_subset = `FALSE`,
n_excluded_subset = `TRUE`)),
# Self reported evaluations
# US_contingency_memory
(combined_data %>%
# EXCLUDE CONDITIONS THAT HAD INADEQUATE N TO CALCULATE ES IN METAS ABOVE
filter((!experiment %in% c("8nri") & US_contingency_memory == TRUE) |
US_contingency_memory == FALSE) %>%
select(EC_Effect_CS1_CS2_Difference_Score, source_valence_condition, US_contingency_memory) %>%
drop_na() %>%
count(US_contingency_memory) %>%
mutate(moderator = "US_contingency_memory",
DV = "self reported evaluations") %>%
spread(US_contingency_memory, n) %>%
rename(n_robustness_subset = `TRUE`,
n_excluded_subset = `FALSE`)),
# color_valence_contingency_memory
(combined_data %>%
select(EC_Effect_CS1_CS2_Difference_Score, source_valence_condition, color_valence_contingency_memory) %>%
drop_na() %>%
count(color_valence_contingency_memory) %>%
mutate(moderator = "color_valence_contingency_memory",
DV = "self reported evaluations") %>%
spread(color_valence_contingency_memory, n) %>%
rename(n_robustness_subset = `TRUE`,
n_excluded_subset = `FALSE`)),
# demand_self_reports
(combined_data %>%
select(EC_Effect_CS1_CS2_Difference_Score, source_valence_condition, demand_self_reports) %>%
drop_na() %>%
count(demand_self_reports) %>%
mutate(moderator = "demand",
DV = "self reported evaluations") %>%
spread(demand_self_reports, n) %>%
rename(n_robustness_subset = `FALSE`,
n_excluded_subset = `TRUE`)),
# reactance_self_reports
(combined_data %>%
select(EC_Effect_CS1_CS2_Difference_Score, source_valence_condition, reactance_self_reports) %>%
drop_na() %>%
count(reactance_self_reports) %>%
mutate(moderator = "reactance",
DV = "self reported evaluations") %>%
spread(reactance_self_reports, n) %>%
rename(n_robustness_subset = `FALSE`,
n_excluded_subset = `TRUE`)),
# behavioural intentions
# US_contingency_memory
(combined_data %>%
select(intention, source_valence_condition, US_contingency_memory) %>%
drop_na() %>%
count(US_contingency_memory) %>%
mutate(moderator = "US_contingency_memory",
DV = "behavioural intentions") %>%
spread(US_contingency_memory, n) %>%
rename(n_robustness_subset = `TRUE`,
n_excluded_subset = `FALSE`)),
# color_valence_contingency_memory
(combined_data %>%
select(intention, source_valence_condition, color_valence_contingency_memory) %>%
drop_na() %>%
count(color_valence_contingency_memory) %>%
mutate(moderator = "color_valence_contingency_memory",
DV = "behavioural intentions") %>%
spread(color_valence_contingency_memory, n) %>%
rename(n_robustness_subset = `TRUE`,
n_excluded_subset = `FALSE`))
) %>%
select(DV, moderator, n_robustness_subset, n_excluded_subset)
# print
n_per_cell %>%
kable() %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = FALSE)| DV | moderator | n_robustness_subset | n_excluded_subset |
|---|---|---|---|
| IAT | US_contingency_memory | 484 | 149 |
| IAT | color_valence_contingency_memory | 348 | 218 |
| IAT | demand | 854 | 75 |
| IAT | reactance | 839 | 96 |
| self reported evaluations | US_contingency_memory | 865 | 256 |
| self reported evaluations | color_valence_contingency_memory | 625 | 429 |
| self reported evaluations | demand | 1211 | 212 |
| self reported evaluations | reactance | 1222 | 201 |
| behavioural intentions | US_contingency_memory | 805 | 249 |
| behavioural intentions | color_valence_contingency_memory | 625 | 429 |